Conscious AI? FUnctionalism and its Alternatives

We are happy to announce that the 13th annual Ernst Mach Workshop, entitled Conscious AI? Functionalism and its Alternatives, will take place in Prague, on June 2–3, 2025. The workshop, organized by The Institute of Philosophy, Czech Academy of Sciences (www.flu.cas.cz) in cooperation with PAIR Erlangen (https://www.pair.fau.eu/), will focus on the question of whether an artificial, e.g. silicon-based, system could be phenomenally conscious, and more generally, on the relation(s) between artificial intelligence (AI) and (conscious) mentality. Our keynote speakers will be:

• Anil Seth (University of Sussex)
• Kathinka Evers (Uppsala University)

The recent rapid progress in the field of deep learning has led to the creation of AI systems that exhibit abilities once thought to be reserved for humans. While primarily trained for text and pattern prediction, Large Language Models (LLMs) and other generative AI models, have been increasingly able to converse with their users in sophisticated ways and perform remarkably well on many, often multimodal tasks that, until recently, were taken to require human-level intelligence. The stunning skills exhibited by such AI systems have significantly revived researchers’ interest in the question of whether an AI system could be conscious in the sense of there being something it’s like subjectively for this system to be this system. At our workshop, we will investigate the main reasons for thinking that current or future AI systems could be conscious as well as the main arguments against this suggestion. We aim to examine the following questions, among others:

•  What is the best method of determining whether an AI system is, or could be, conscious?
• Are present or future GenAI models prime contenders for AI consciousness, or are there other AI systems that may be better suited for consciousness?
• What are the main obstacles on the path to AI consciousness (if there are such obstacles)?
• Proponents of the possibility of AI consciousness often presuppose that computational functionalism is true. How plausible is this presupposition? Could one have AI consciousness without computational functionalism?
• If AI consciousness is possible, how similar could it be to human consciousness?
•  Is it plausible that present or future implementations of generative AI might be able to think, or understand the texts and other inputs they process?
• What does the uncertainty regarding consciousness of (future) AI systems mean from the ethical point of view? Are there societal risks associated with creating conscious AI?

Presentations on other, closely related topics, relevant to the research interests of our keynote speakers, are also welcome. The workshop is intended to be in-person and discussion-oriented – presenters of contributed papers are given time slots of 45 minutes (including discussion).

If you’re interested in participating in the workshop, please send an anonymized 300-word abstract of your paper to [email protected] by February 28, 2025. (This e-mail address can also be used for enquiries regarding the workshop).

Notification of acceptance: end of March, 2025

There is no registration fee for the workshop.